Modeling and utilizing habits using process mining for building spatial design systems

Residents need to change their habitual behaviors following living space changes, such as moving or remodeling, and that may occur mental stress. This stress is a major problem, especially for the elderly, who are less able to cope with changes in their environment. To reduce this stress, a system that reflects the living information of the original houses in new houses, where habits can be retained in the new environments is needed. Many studies have been conducted to quantify life information as a habitual model using data mining and pattern recognition methods. “Process Mining” is a theory developed to visualize and improve processes in the business field and applied to lifestyle information, and it is possible to create a habit model. In recent years, several studies on habit models using process mining have been reported. However, there are no studies in which these process mining-based habit models have been adopted to design architectural spaces such as living spaces. Therefore, the purpose of this study is to investigate the relationship between habit and architectural space by utilizing a process mining-based habit model. Specifically, we propose the automatic extraction and visualization of habit behaviors through process mining and the use of habit models. The data acquisition experiment was conducted in an experimental smart home. This smart home is a mobile trailer house built by a multi-company project and is equipped with many sensors that can automatically acquire many daily living data. Subjects were recruited randomly and lived alone in this smart home for one week. An input matrix was created from the acquired data set and process mining was adapted to create habit models. In this study, two habit models were created: (1) a habit model based on behavioral information and (2) a habit model based on location information. Each input matrix consisted of (1) 16 types of behavior record data manually entered by the subject and (2) ground reaction force data in the house divided into 7 areas. We investigated the relationship between habitual behaviors and spatial conditions by integrating these two models.